+ All Categories
Home > Documents > Warlike Democracies

Warlike Democracies

Date post: 15-Nov-2015
Category:
Upload: anonymous-qvdxo5xtr
View: 8 times
Download: 0 times
Share this document with a friend
Description:
War
Popular Tags:
37
Warlike Democracies Author(s): John Ferejohn and Frances McCall Rosenbluth Source: The Journal of Conflict Resolution, Vol. 52, No. 1 (Feb., 2008), pp. 3-38 Published by: Sage Publications, Inc. Stable URL: http://www.jstor.org/stable/27638593 . Accessed: 27/09/2013 11:18 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Conflict Resolution. http://www.jstor.org This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AM All use subject to JSTOR Terms and Conditions
Transcript
  • Warlike DemocraciesAuthor(s): John Ferejohn and Frances McCall RosenbluthSource: The Journal of Conflict Resolution, Vol. 52, No. 1 (Feb., 2008), pp. 3-38Published by: Sage Publications, Inc.Stable URL: http://www.jstor.org/stable/27638593 .Accessed: 27/09/2013 11:18

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    .

    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

    .

    Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal ofConflict Resolution.

    http://www.jstor.org

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Warlike Democracies John Ferejohn Department of Political Science, Stanford University, Stanford, California Frances McCall Rosenbluth

    Department of Political Science, Yale University, New Haven, Connecticut

    Classical republican theories are monadic in the sense of seeing in each political regime a set of typical operating characteristics. There is disagreement as to what those charac

    teristics are and specifically whether republican governments are more likely to be agg ressive or peace loving. We group these two views as (democratic) mobilization theory versus (republican) checks theory and argue, first, that each can help us understand the

    finer structure of republican government; second, that they are not contradictory but can

    be combined in various ways in the same institutions; and third, that they offer the pro

    spect of deepening our understanding of what is called the democratic peace proposition.

    Keywords: Machiavelli; Kant; democratic peace; mobilization

    1. Introduction

    Long before the current "age of democracy," political commentators from Thu

    cydides and Polybius to Machiavelli and Kant advanced arguments about the

    Authors' Note: We thank Bruce Russett for his generosity in sharing data from his research on the Pelo

    ponnesian War, and for helpful comments on several versions of this paper. We also thank Emily Mackil, Ian Morris, Josh Ober, Pasquale Pasquino, Walter Scheidel, and Barry Strauss for participating in a workshop on War and Politics in Ancient Greece on December 5-6, 2004; Mary Beard, Tim Cornell,

    William Harris, Andrew Lintott, Wilfried Nippel, and Nate Rosenstein for a workshop on War and Poli

    tics in Republican Rome on March 20-21, 2005; and Bill Caferro, Sam Cohn, Steven Epstein, Pasquale

    Pasquino, Christine Shaw, and David Wootton for a workshop on War and Politics in Medieval and

    Early Modern Italian City Italian city-states on December 14-15, 2005. We are grateful to the Bellagio Study and Conference Center of the Rockefeller Foundation for supporting the Italian city-states work

    shop and to the Georg W. Leitner Program and the Yale Center for International and Area Studies for

    financial support that made the workshops possible. Alisa Ardito, Michelle Tolman-Clarke, Curtis

    Eastin, Erica Franklin, Bryan Gervais, Harris Mylonas, and Sandy Henderson provided able research

    assistance, and Mario Chacon was an excellent statistical assistant and consultant. We also thank

    Kenneth Arrow, Gary Cox, Robert Dahl, Keith Darden, Shigeo Hirano, Istvan Hont, Bob Keohane,

    Joseph LaPalombara, John McCormick, Kevin Quinn, Dan Reiter, Chuck Sabel, Kenneth Schultz, Ian

    Shapiro, Jim Vreeland, (dear, late) Michael Wallerstein, Barry Weingast, David Weinstein, and partici pants in the Political Economy Colloquium at the Hoover Institution in 2006 for helpful comments. Data are available in the appendices, and in digital form at http://jcr.sagepub.com/supplemental.

    Journal of Conflict Resolution Volume 52 Number 1

    February 2008 3-38 ? 2008 Sage Publications

    10.1177/0022002707308596

    http://jcr.sagepub.com hosted at

    http://online.sagepub.com

    3

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 4 Journal of Conflict Resolution

    qualities of "republican" government that might affect their war-making capacity and posture relative to those of autocratic or oligarchic polities. Their arguments are monadic, in the sense that they describe attributes of republics that are thought to affect their foreign policy behavior, regardless of the circumstances in which

    they find themselves or the other polities with which they interact. Modern scholar

    ship on democratic foreign policy has lost touch with some of this early thinking, particularly the Machiavellian view of republican warlikeness, but it has ignored the monadic elements of Kant's theory as well. We argue that, in fact, modern

    democratic republics have attributes that cut in different directions. This helps make sense of the seemingly contradictory findings from the democratic peace literature that democracies tend to forbear against each other while making effec tive use of muscle elsewhere (Lake 1992; Reiter and Stam 2002; Biddle and Long 2004; Schultz and Weingast 2003).l

    In this article, we evaluate theories of democracy and cluster them into two

    groups, the first of which we call Kantian, and the second, Machiavellian. We derive these labels from the writings of Kant and Machiavelli, of course, but the labels are meant to capture a stylization of central arguments of these great thinkers

    rather than to provide a full exegetical account of their work. We take Kantian institutions to stand roughly for the idea of checked or mixed governmental power that, among other things, could inject caution or deliberateness into foreign policy decision making. Machiavellian or neoclassical theory stands for the idea that enfranchisement may increase the likelihood of winning wars by drawing more

    fully on societal resources. In the end, we conclude that neither variety of monadic

    democracy theory provides conclusive foreign policy predictions because modern democracies combine elements of both Kantian and Machiavellian logics. Indeed, turning the causal logic around, we might speculate that the standard shape of the

    modern democratic republic, which combines mass franchise with institutional checks, arose in the context of centuries of interstate competition in which well

    mobilized but duly deliberate societies enjoyed a survival advantage. The scholarly literature has begun to point out that "democracy" is a complex

    concept that agglomerates a range of institutional and other features with separate effects on foreign policy decision making (Prins and Sprecher 1999; Clark 2000; Reiter and Tillman 2002; Chan and Leblang 2003; Howell and Pevehouse 2007). In this article, we join this endeavor to unpack the institutional bases of political regimes, focusing on two dimensions that can be compared across wide swaths of

    time and place: "Kantian" institutional checks, which count the number of collec

    tive veto players in a political system; and "Machiavellian" political inclusivity, which we define as the effective franchise, or the proportion of the population that can influence decisions about war and peace and the distribution of the spoils and costs of war. As we elaborate in the following section, institutional checks include not only horizontal checks or "separation of powers," but also what we call a verti

    cal check, or representation, allowing us to distinguish between direct and indirect

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 5

    democracies. We argue that both kinds of "republican" checks instill caution in

    foreign policy and work in the opposite direction from political inclusivity that Machiavelli theorized to enhance war-fighting capacity.2

    We therefore separate democracy in the strict sense?typified by the Athenian form of direct rule by popular assemblies and lottery-selected magistrates?from modern democratic states, which exhibit vertical and sometimes horizontal checks and a wide franchise. The Athenian state was based on a fairly small franchise by modern standards but also contained few institutional checks on what the popular assembly could do in matters of peace and war.3 Republican Rome had a wider franchise but also more checks. Modern democratic states differ similarly: the United States since the Civil Rights era has a wide franchise with many checks, while modern Britain has a wide franchise with few checks. On the issue of how the popular influence is expressed, Athens and Rome are on one side, with direct

    popular influence expressed in popular assemblies; and Britain and the United States are on the other, with popular influence working through representatives, which Kant regarded as perhaps the most effective kind of check on popular impulses. By differentiating among political institutions along these dimensions,

    we are able to theorize about and test these attributes separately, at least in princi

    ple. Rather than seeing a single effect of democratic republics, we argue that politi cal regime characteristics can generate countervailing effects. Moreover, in

    countries that have these attributes in different combinations, we expect to see dif

    ferent patterns of outcomes. We contribute to the democratic peace literature not

    by proposing a novel explanation of the tendency for democracies to avoid fighting each other, but by parsing the effects of institutional forms to suggest the condi tions under which democratic republics are likely to be expansionist.

    We distinguish our approach from the mainstream democratic peace literature in two other ways. Methodologically, we question the sufficiency of empirical stra

    tegies that focus on what amount to long-run differences between states, obscuring

    intertemporal or short-run changes in regime types.4 At least some of the variation

    in those studies that pool cross-sectional and time series information traceable

    to complex differences between states and many important variables that might account for cross-sectional variation are unmeasured. Moreover, by focusing on a

    truncated time period beginning after the Napoleonic Wars, much of the quantita tive democratic peace literature suppresses variation that would appear in a larger time frame such as the rise of the mass army from the fifteenth century onwards,

    and the existence of more varied regimes.5

    Second, we think that expansionary success rather than war itself might be a more appropriate dependent variable for measuring assertiveness.6 As many have

    noted, fighting is endogenous: sometimes an aggressive policy that claims land and resources from others will be resisted violently but usually only if the r?sister thinks it stands some reasonable chance of success, however that may be defined.7

    Whether a nation will rationally decide to fight depends on the beliefs of the actual

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 6 Journal of Conflict Resolution

    potential combatants as to their relative capacities and on their assessments of

    what is at stake, and these capacities and valuations will reflect things that are irre

    levant to our theoretical interests. Resource advantages, technical progress, popula tion size, geographical circumstances, whether one party or the other has exhausted

    itself in other wars, or whatever. We want to put all of these things to one side.

    The rest of the article is organized as follows. Section 2 sets out Kant's thinking hypothesis as to how republican institutions may reduce the inclination of self

    governing populations to wage war. We will see there that his ideas as to why cer tain regimes would be peace loving were more nuanced and complex than is com

    monly recognized. Section 3 considers Machiavelli's contrasting argument about

    democratic capacity and aggression. While Kant stresses the way different regimes weigh the costs of war, Machiavelli notes that democratic publics may favor wars

    they think they can win because they expect to share in the spoils of victory. Taken

    together, the theorists provide a template with which to understand the foreign policy behavior of a range of regime types.

    Section 4 lays out longitudinal evidence that can better capture the unit-level sources of foreign policy choices than can aggregate, cross-sectional data. The his

    torical record, though only illustrative, suggests to us that superior military capacity may override reasons for caution that self-rule generates. Section 5 makes use of a

    period of Italian history between 1200 and 1600 in which there was substantial institutional variation among neighboring states to examine the effects of institu tional checks and of popular voice on the territorial reach of Italian city-states.8 Section 6 concludes with the observation that while checks may restrain govern ments by injecting the decision-making process with a supermajority requirement, the "democratic peace" of the past century may be more the result of democratic

    deterrence and liberal institutions than of a fundamentally peaceful orientation of modern democracies.

    2. Kant and Neo-Kantianism

    The father of the democratic peace idea, Immanuel Kant, argued in a familiar and much quoted passage, "If (as must inevitably be the case, given this form of

    constitution) the consent of the citizenry is required in order to determine whether or not there will be war, it is natural that they consider all its calamities before

    committing themselves to so risky a game_By contrast, under a nonrepublican constitution where subjects are not citizens, the easiest thing in the world is to declare war" (Perpetual Peace, as quoted in Reiss 1991, 113). Citizens who could decide for themselves whether to wage war would be more conservative than a

    head of state who could better insulate himself from the costs and miseries of war. An autocrat, by contrast, "is not a fellow citizen but the owner of the state, and a

    war will not force him to make the slightest sacrifice so far as his banquets, hunts, pleasure palaces and court festivals are concerned. He can thus decide on war,

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 7

    without any significant reason, as a kind of amusement, and unconcernedly leave it

    to the diplomatic court... to justify the war for the sake of propriety" (ibid., in Reiss 1991, 100). The sense of this is clear enough: a republic (if it has a wide fran

    chise), because everyone (or perhaps every male) is a citizen, will internalize the costs of war and nonrepublican governments will not.9

    Immediately after the passages stating the republican peace proposition, Kant went on to distinguish republican from democratic government: "Republicanism is that political principle whereby executive power is separated from legislative power. In a despotism the ruler independently executes the policy that it has itself made ... here rulers have taken hold of the public will and treated it as their own private will. Among the... forms of government, democracy, in the proper sense of the term, is necessarily a despotism..." (ibid., in Reiss 1991, 114). Like

    Montesquieu, therefore, Kant distinguished despotic from republican government and the key distinction for both thinkers is that a republican government rules

    through law and not arbitrarily. Kant thought republics differed from democracies

    (and despotisms generally) in two ways. First, republics separate legislative and executive powers. The body authorizing action is different from the body taking the action; we call this a horizontal separation of powers. Second, Kant also says that in republics ruling is done by representatives and not directly by the people.10 This creates a vertical separation by which representatives rather than the people take authoritative actions. For Kant, the people have no direct role in republican

    government as they had in both the Athenian ekklesia and the Roman comitia.

    Quite clearly, Kant thought that his peace proposition applied specifically to

    republics rather than to democracies. In this sense, his view was not in conflict with

    the classical belief that democracies of the classical or direct kind are especially warlike. The Greeks would have thought exactly the same thing: they thought that

    republican cities like Sparta were less likely to be militarily adventurous than democratic Athens. Direct democracies were likely to be turbulent and swayed by emotion rather than reason. For this reason, Kant thought democracies despotic (or

    will-driven) in that they could be moved to take action based on momentary pas sions. Republics, because they had extensive horizontal and vertical checks, would be more likely to consider dispassionately any proposal to fight.

    In a move that has become central to the "liberal" variant of democratic peace

    theory, Kant went on to amend his monadic theory with dyadic logic. Kant thought it unwise even for republics to rely solely on the first definitive article of perpetual peace to prevent war, for he added second (international organizations) and third

    (universal hospitality, or trade) d?finitive articles. While Kant may have thought it

    possible for humans' gravitation toward republican self-rule to take care of much of the world's ills, a better world might arise in this way only after "much inconve nience" (Reiss 1991, 113).

    Kant's prescription for perpetual peace has found prolific modern-day expres sion in scholarship advancing "the democratic peace proposition."11 And partly

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 8 Journal of Conflict Resolution

    for this reason, most current writers think of the democratic peace as a property

    of dyads?pairs of democracies. Many contemporary theorists have dropped or

    de-emphasized Kant's sharp monadic distinction between democracies and republics, taking instead the broad and modern definition of a democracy as a self-governing polity of more or less politically equal citizens. We think this is a potentially impor tant distinction to revive, at least theoretically, because decision-making checks and

    political inclusion are logically and empirically independent to at least a certain

    degree and should exert separate influences on foreign policy.

    2.1. Direct and Representative Democracies

    For classical theorists, on whom Kant relied extensively for his constitutional

    classifications, monarchy, aristocracy, and democracy could exist in their pure forms

    or could be combined in various ways to form "mixed government." Aristotle's pre

    ferred system of government was a democracy with a sufficiently strong property qualification that would permit the middle class to rule, but he also thought that a mixed constitution could achieve a similar effect if it could balance the powers of the multitudinous poor and the few who are rich: "The farther away one moves from

    middle-polity, the worse"; and "When the unpropertied class without the support of a middle class gets on top by weight of numbers, things go badly and they soon come to grief" (Aristotle 1960, IV. 11). Aristotle's preference for a either a high property qualification or a mixed constitution rested on his view that unchecked majoritarian institutions would fail to give sufficient weight to the middle class of property own

    ers, which "is also the steadiest element, the least eager for change. They neither

    covet, like the poor, the possessions of others, nor do others covet theirs, as the poor

    covet those of the rich. So they live less risky lives, not scheming and not being schemed against" (ibid.). For Aristotle, a system of government that privileged ele

    ments of society with a propertied stake, by way of magistracies or leadership coun cils of one kind or other, would likely make superior decisions about the use of

    community resources in war and peace.12 While Kant adopted the classical threefold typology of constitutional types and

    shared Aristotle's disdain for direct democracy, his preoccupation was with how to achieve rule of law rather than with moderate policies. For Kant, direct democracy would be as arbitrary and fickle as despotism, even if its politics were usually moderate, because it had no way to commit itself to a rule of law that would govern impartially all members of society, including the existing majority. Kant's pre ferred solution was a representative government that would combine the merits of

    allowing people to be self-governing while stabilizing the laws by making them hard to change in application. Kant was not much clearer than this, but it seems

    important to consider the mechanisms that he thought would make indirect or

    representative democracy conducive to rule of law. He thought this would work by introducing two kinds of checks on government.

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 9

    Horizontal Checks

    In one passage, Kant equates direct democracy with despotism for fusing the

    legislative and executive functions of government. By contrast, "republicanism is

    that political principle whereby the executive power (the government) is separated from the legislative power" (as quoted in Reiss 1991, 101). Here he is making a case for lateral checks by one branch on another to restrain an impetuous majority. Kant is in good company in making this claim, for it draws on a rich constitutional tradition from Aristotle and Polybius to Machiavelli and Montesquieu. We have

    already recounted Aristotle's low opinion of majoritarian politics. Polybius thought of Rome that "the proper equilibrium is maintained by the impulsiveness of the one part being checked by its fear of the other," referring to the Senatorial and

    popular elements of government balancing each other (Polybius 1962). Machiavelli echoed this position, stating "all legislation favorable to liberty is brought about the clash between them [referring to the popular and aristocratic elements of gov

    ernment]" (Machiavelli 1674, 1.4). In modern terminology, checks and balances can increase the commitment to

    rule of law by increasing the transactions costs of changing laws. Whether or not veto points are lodged in particular sociological groups with antithetical interests, as ancient writers conceived, any check is likely to slow down the process of policy change because no two aggregations of citizens' interests are likely to be exactly the same. As we know from spatial theory, changing policy becomes harder the farther apart the ideal points of the veto holders, and this in turn depends on the rules by which the veto players in possession of the checks are chosen.

    While lateral checks may slow down the process or increase the costs of policy change, we can see from modern examples that their absence does not necessarily eliminate the possibility of rule of law, as a strict reading of Kant's text implies. Parliamentary sovereignty systems, which deliberately fuse the legislative and executive functions of government, may indeed enable legislative majorities to

    change laws with greater ease than their counterparts in checked systems. But

    Kant thought that a second, vertical, set of checks built into representative systems qualified parliamentary sovereignty systems as republics and spared them from the "despotism" of direct democracies.

    Vertical Checks

    Representative governments, by definition, separate decision making from the

    people. In most matters, the public delegates to their representatives decision

    making authority to act on their behalf and thereby reduce the costs of decision

    making. Kant actually suggests that complete delegation is not possible in serious matters such as a declaration of war, where the people must somehow consent to

    these decisions directly by agreeing to fight. A vertical check thereby arises in these

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 10 Journal of Conflict Resolution

    instances when the indirect aspect of representation is somehow incomplete: a resi

    dual undelegated power remains with the citizenry. Vertical checks of this sort would work in much the same way as lateral checks

    and balances, by raising the threshold for making and changing laws. But in one

    way they are fundamentally different: vertical checks work to limit democracy directly or, if you prefer, to limit direct democracy. Of course, in the event that the distribution of the legislative assembly's preferences over going to war turned out to be identical to that of the voting public, this check would lack force. But as Kant

    emphasized, when it comes to decisions to go to war, the preferences of representa tives are likely to differ systematically from those they represent, so convergence is

    unlikely. Kant is not likely to have thought, nor do we think, that checks slow down

    policy making in all circumstances, but institutional checks are more likely to con strain as the utility of aggression becomes more ambiguous or contested. In clear

    cases of self-defense, even a regime with multiple veto players would likely reach

    swift consensus on the appropriate policy response, such as the United States after

    Pearl Harbor, for example. Checks potentially induce prudence (or at least generate the time for prudence to become manifest) and reduce the chance of passionate or rash action by requiring more institutional actors to agree to the net expected costs.

    It is also possible theoretically to think about a gradation along a republican dimen

    sion, although measurement issues are potentially thorny.13

    3. War Capacities and Constitutions

    3.1. Democratic Capacity

    Machiavelli approached the problem of warlike democracies with radically dif ferent motives from Kant's.14 Wishing to find or create for his native Florence some of Republican Rome's vigor and power in order to resist Spanish, French, Austrian, and papal influence, Machiavelli concluded that politically inclusive

    regimes were more likely to be successful at war and therefore to be expansionist because their more thorough mobilization of the populace would give them a mili

    tary advantage over poorly mobilized neighbors (Machiavelli 1674, 1.6). For

    Machiavelli, it was precisely because Republican Rome gave its plebeian soldiers

    political voice, and because it enfranchised the populations of the lands it con

    quered, that it was militarily powerful, and therefore successfully republican for a

    long period of time. A broader political franchise gave Rome more citizen-soldiers, and they were all fighting for a system in which they had a stake. Machiavelli in fact thought that a well-organized republic must be militarily expansive.15

    A Machiavellian explanation for why politically inclusive regimes were milita

    rily expansive rests on the public's motivation to fight wars they could win, since

    they would benefit from the gains of victory as surely as they would bear the costs

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 11

    of fighting. Machiavelli was contrasting the stronger motivation of soldiers to fight for a polity in which they possessed a voice with the disinclination of oligarchic governments to arm subjects who might use arms to wrest political concessions.16

    History bears him out on both counts. Oligarchical governments are famous for

    building their militaries around the mounted nobility, fearful of arming the peasan try. And sometimes, at the end of the day, they are forced to abandon this policy. Sparta had little choice but to rely on its slaves and subject allies when its own

    population became insufficient to fend off military threats, though it reserved key battlefield roles for its own troops. In any case, the new policy did not work. In 371 b.c. the Theban general Epaminondas defeated the Spartan army at Leuctra by concentrating the force of his assault against the Spartan hoplites rather than going after their allies first. As he expected, the slave and allied flanks gave way as soon as the Spartan ranks fell. Heavy French losses to England at Agincourt in 1415 are often attributed to the unwillingness of the French nobility to arm peasants with crossbows (Landers 2003), and the dismemberment of Poland by Prussia and Russia in 1795 is arguably attributable to a similar fear of placing lethal weapons in the hands of serfs. In his own land, Machiavelli thought the weakness of the Italian states, which relied heavily on mercenary troops rather than recruiting citi

    zen armies, were too weak to withstand the powerful European states.

    3.2. Republican Capacity

    Since at least the Middle Ages, republican regimes?defined as regimes with checks or by extension as regimes that rule by law?have an additional mobiliza

    tional advantage in being able to borrow funds cheaply and in enforcing property rights necessary to foster economic growth. As North and Weingast (1989) argued in the case of the Glorious Revolution in England in 1688, making the king subject to a parliamentary check and strengthening judicial oversight assured investors that the monarchy would no longer be able to renege on loans or to confiscate other

    property. By their telling, England's military might grew substantially in compari son to that of France where the monarch could not credibly commit to honoring contracts to which he was party. The higher cost of French debt and France's weaker economic performance were, in part, constitutionally determined.

    Nondemocracies, to be sure, have also extracted heroic sacrifices from their sub

    ject populations, and repressive regimes of all stripes have employed some combi nation of fear, nationalism, religion, and ideology with considerable success.17 We

    do not mean to suggest that democracy is the "magic bullet," so to speak. The

    Russian empire, for example, was not particularly inclusive, but it expanded stea

    dily from the 1500s for several centuries.18 At first much of this expansion was into the steppe lands and tundra that went relatively uncontested. Once Russia gained a certain scale, it became difficult for smaller entities, however well mobilized on a

    per capita basis, to challenge the encroachment effectively. The more inclusive yet

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 12 Journal of Conflict Resolution

    substantially smaller Swedish state shrank back rather than take on the enormous Russian army. And as Napoleon's Grand Armee found when it marched toward Moscow in July 1812, taking only summer uniforms in confidence of swift victory, the Russian territory was so vast that the Russian armies were able to draw the

    French ever deeper without directly engaging Napoleon. It was the Russian winter more than its men that subsequently defeated Napoleon's army, leaving alive a

    small fraction of the French forces to return, humiliated, to Paris.

    History gives us a fair number of examples of repressive empires, and of politi cal leaders who managed to mix selective compensation with fear and other ele

    ments of the human psyche to generate strong apparent support for their tyranny. We do not claim that only democracies have the sort of capacity advantage that allows them to expand. Our point is more modest: that politically inclusive regimes are on balance likely to enjoy higher levels of soldier morale and initiative, popular support, voluntary tax compliance, economic vibrancy, and public involvement in

    the choice of wars, all of which increase the odds of success to one degree or another.

    3.3. Machiavelli's Democratic Peace

    If democratic republics are better at mobilizing manpower, information, and

    other resources for the wars they choose to fight than are other regimes, this advan

    tage would be expected, ceteris paribus, to shift the distribution of costs and bene fits in their favor. If this is so, it may be the case that democratic republics are more

    war-prone (as Machiavelli thought) simply because they are better positioned mili

    tarily than otherwise identically situated nations.19 This effect might be especially potent in situations in which there are relatively few democracies: in that context, a democratic republic is likely to face an especially tempting set of targets for

    aggression.

    By Machiavelli's logic, any particular popular regime's mobilizational advan

    tage would disappear when faced with other similarly mobilized regimes, all else

    equal, and it would adjust its military posture accordingly. Democratic republics should be more peaceful than other regimes only when their expected costs from

    fighting outweigh their expected benefits. As with any effective deterrence, wide

    spread democratic mobilization should generate peaceful interstate relations by reducing the expected benefits from war.

    4. Some Evidence

    In this section we explore, with historical examples, the separate effects of fran

    chise and institutional checks. The number of cases we are able to treat in detail is

    obviously too limited to make statistically defensible claims; our aim here is rather

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 13

    to investigate the empirical plausibility of the institutional effects we have exam ined theoretically. We are aware of enormous problems of case selection, con

    strained as we are by the paucity of data on a hypothetically fuller set of cases. But

    by examining several regimes that underwent substantial institutional change over

    time, we are able to exploit natural experiments in which to observe attending shifts in foreign policy.

    We choose our cases on the basis of variation along the two independent dimen

    sions of interest, political inclusivity and checks (see Figure 1). We expect regimes with a large franchise and few checks (Athens and post-Marian Rome, for example) to be the most expansive. Highly checked regimes with a narrow franchise (Sparta,

    medieval Italian city-states, early modern Poland?all of which were republics in some sense) ought to be the most militarily reluctant. Narrowly based regimes with few checks, such as autocracies, and widely based regimes with multiple checks, such as the United States, are cross-pressured. A third dimension, relative size and

    therefore potential military resources, may of course overwhelm the first two.

    By including in our analysis territorial expansion and foreign policy assertiveness rather than focusing on wars per se, we steer clear of the trap of considering only cases where informed calculation broke down.

    Expansionist Powers: Large Franchise, Few Checks

    The world has documented few direct democracies, but the classic case, of course, is Athens in its prime. Republican Rome also had relatively broadly based

    popular elements compared to other republics, particularly in the late republic when checks began to break down, which might account for the pronounced influ ence of populist leaders like the Gracchi brothers or Marius or Caesar. We could also include nineteenth-century Britain, which simultaneously expanded its fran

    chise and reduced its institutional checks. By our reckoning, these regimes should have been expansionist powers, and indeed they were.

    Athens. Of course, neither Machiavelli nor Kant would have expected Athens of the fifth century b.c. to be militarily timid, Machiavelli because Athens was sur rounded by regimes less inclusive than itself that were easy pickings, and Kant because Athens was a radical democracy in its decision-making system. Indeed,

    Athenian aggressiveness is well known. At the peak of its fifth-century empire, Athens built a massive fleet paid in large part by the tribute it extracted from the members of its alliance structure, with which it dominated commerce and protected its colonies and allies in the Aegean. The Aristotelian author of the Constitution of the Athenians pointed out that at least some of the tribute money was spent domes

    tically on payment for public service in juries and the boule as well as the public building program and for poor relief (Constitution of the Athenians, 24/2, as cited in Samons 1998, 44). Plutarch has Pericles saying, in about 450, that the Athenians

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Figure 1 Franchise and Checks in Historical Cases

    Franchise

    Low High

    Low

    Checks

    Autocracies Direct Democracies (Athens, 5th

    century B.C.)

    Post-Marian Roman Republic

    19th Century Britain

    High Sparta

    Medieval Venice

    Early Modern Poland

    U.S.

    -PI'

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 15

    were not obliged to give the allies any account of how their money was spent provided that they carried on the war for them and kept the Persians away (Plutarch 1960, 12). Although a powerful protector, Athens was not necessarily the object of the fondness and affection of its tribute states (Thucydides 1954, 75; though Ste. Croix [1955] argued that Thucydides exaggerated the ill will Athens brought on

    itself). Some notorious examples of Athenian rashness may seem to bear out the Kan

    tian interpretation that imprudence was partly associated with Athenian aggrand izement: in 428 the Athenian assembly voted to slaughter all Mytilenean males for revolting against the Athenian alliance?though a subsequent assembly vote reversed the decision, requiring the dispatch of fast rowers to intercept the first

    boat; they authorized the Sicilian expedition in 421 on Alcibiades' urging without

    considering what they were up against;20 and they voted to execute their (victor ious!) generals for failing to pick up dead or possibly drowning sailors in stormy weather following the battle of Arginousai in 406. Note, however, that the rashness on these occasions may have contributed to war but not to successful expansion. The Sicilian expedition was a disastrous failure, and executing their best generals paved the way for the crushing defeat, with inexperienced generals at the helm, at

    Aegospotamai in 404. How Athens would have behaved had it been constrained by Kantian institutional

    checks takes us into the realm of speculation. Already in the fifth century the Athe nians had a lottery-chosen council (boule) that set the agenda for the assembly, annually elected generals, and checks on leaders in the form of impeachment and

    postterm examination backed by possible sanctions and even ostracism.21 When

    democracy was restored in 403 after the oligarchic interval, the Athenians instituted the nomothetai, a legislative/judicial board charged with the solemn responsibility of passing or rejecting any law proposed by the assembly (Hansen 1991, 163).

    After Thebes defeated Sparta at Leuktra in 371, Athens had difficulty securing tribute from other Greek poleis because Athens no longer had an undominated

    reputation with which to sell protection to surrounding city-states, and perhaps also because it had become a more cautious polity (McCoy 1991; Perlman 1968; Seager 1967; Cawkwell 1997, 41; Habicht 1996). Gone also was a portion of Athens's relative capacity advantage, partly because more poleis had inclusive

    regimes in the fourth century than in the fifth (including some of those renegades), and also because a number of poleis in Boeotia (led by Thebes), and later Aetolia, and Achaea formed their own defensive leagues among themselves as an alterna

    tive way of marshalling sizeable armies (Mackil 2004; Botsford 1910; Bakhuizen

    1994; Larsen 1952).22 Additional layers of judicial oversight may have stabilized the democracy in fourth-century Athens, contributing to a flourishing economy and commercial success (Fine 1983). But surrounded by more fully mobilized regimes,

    Athens was not able to translate this economic vibrancy into a reformed empire. The days of getting neighbors to pay for the Athenian fleet were largely over.

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 16 Journal of Conflict Resolution

    The Roman Republic. Institutional form is perhaps even more important for

    explaining Rome's expansionist push beyond Italy in the late Republic. Machia velli admired the Republic's mixed or "Polybian" political system, in which demo cratic, oligarchic, and monarchical elements are balanced, but he believed that the traditional system of checks began to break down in the late Republic under demo cratic pressures.23 These domestic political developments coincided with Rome's

    expansionist surge into the far reaches of Europe and the Mediterranean, a surge fueled by immense new legions. We do not know if Rome was really more territo

    rially ambitious in the late Republican period than earlier, but it was perhaps more successful militarily, and its armies had, necessarily, to become more professional and removed from Roman life more than ever before. This record of success surely

    encouraged additional expansionist projects, which required finding more troops. The Roman Senate, a bastion of the wealthy and conservative, had long sought

    to limit the power of the poor by restricting their ability to serve in the army. In 101b.c, Marius ignored the prohibition on recruiting from the lowest census order, the unpropertied, and raised armies on the largest scale Rome had mustered. The decision to enfranchise the Italians in 87 b.c. further added to the recruitment base for the army. Expansion into Illyria and points east, and upward into Europe, fol lowed under the leadership of a succession of ambitious generals.

    It would misrepresent history to ignore that Rome had already taken control of most of the Mediterranean world while the Republican checks were still in place. In the mid-Republic, annually elected consuls were at the top of the military chain of command and shared executive power with several other groups of magistrates, the Senate allocated military fields of operation to the consuls and controlled access to finances, and the popularly elected tribunes could veto governmental acts and

    presided over the assemblies that could pass legislation. Although decisions to go to war had to surmount institutional obstacles of both the lateral and vertical sort that Kant identified, Rome was in fact constantly at war and exceptionally good at

    it, so perhaps the people felt little need for caution. The broad social base of Rome's military, and the generosity of Rome with its citizenship, gave it access to more fighting men, with a greater stake in the polity and outcome of the fight, than was true for many of the lands that Rome took.

    It is true that Rome embarked on faster and more far-flung expansion for a short time in the dying days of the Republic, but it was a short burst of energy that lasted until a bloody and prolonged civil war consumed Rome's resources and the estab lishment of the imperial throne extinguished the flame. The collapse of the Repub lican institutions was probably ultimately rooted in the stationing of troops overseas for years at a time, where they developed stronger attachments to their

    commanding generals, who looked after them and distributed plunder, than to

    Rome, which was far away. Once in a position of power over their troops, popular

    generals such as Sulla, Pompey, and Caesar defied Republican institutions with

    impunity and eventually seized power for themselves. Ironically, much of the

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 17

    Roman Empire had been built during the Republic, and expanded only incremen

    tally once Roman generals began calling themselves emperor.

    Great Britain. The growth of the British Empire took place during a period when its institutional checks were shrinking and its franchise was expanding. Although Britain had a small franchise when it began accumulating colonial territories in the seventeenth century, British territorial expansion occurred in parallel with the

    expansion of British liberty at home and the disappearance of all but representa tional checks in its domestic political institutions (Woodward 1902, 9). Domestic

    political backing for the British empire grew alongside the size of the British electo rate over the course of the three Reform Acts in the nineteenth century. Kathleen

    Wilson (1988) has argued that the popularity of aggressive foreign policy was partly responsible for the fall of Walpole's relatively pacifist cabinet in 1742. It is also well known that Disraeli's formulation of an electoral platform based on protection of the empire won for the Tories a legislative majority in 1874 and positioned the party to continue to flourish through the third Reform Act (Koebner 1949, 18; Dunbabin 1996, 91; Harcourt 1980), while the Liberals struggled to find an equally popular electoral appeal and were ultimately sidelined as a major party. One junior minister remarked in 1873 that "whatever may be the policy of this or any government, pub lic opinion will not permit the withdrawal of British authority from the west coast of

    Africa" (as cited in Darwin 1997, 623). The British government's assumption of rule in India in the wake of the indigenous revolt against the East India Company in 1857, and the government's tightening of its control of African territories following the European scramble for Africa in the 1880s, did not undermine domestic support for the imperial project. By the time of the Boer War in 1899, "there was not a beg gar in London who did not speak of 'our' rebellious subjects" (Schumpeter 1951/ 1989, 15).

    The public did not use its growing political voice to stop British expansion, per haps because the relative ease with which Britain expanded against soft targets was

    appealing even to those paying taxes and sending family members to military ser vice. As Paul Kennedy (1989, 190) has pointed out, the considerably higher British standard of living relative to its nineteenth-century competitors?France, Russia, and Prussia?reduced the public burden of its defense expenditures. Palmerston's

    motto, "trade without rule where possible, trade with rule where necessary," may have overtaken the more cautious foreign policy inclinations of Walpole precisely because England's growing mobilizational capacity gave it resources to spend and an advantage in competing against less inclusive countries and domains.

    As republican checks more or less melted into parliamentary sovereignty in the nineteenth century, England slowed its imperial drive. This was not because the absence of checks hurt England's investment climate, which was by this time ensured

    by majoritarian electoral competition. Schumpeter (1951/1989, 91) argued that capit alism and commerce eventually undermined empire by making it redundant. While

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 18 Journal of Conflict Resolution

    more compatible with long-run trends than Lenin's opposite claim that imperialism followed necessarily from the advanced stages of capitalism, Schumpeter's expla nation fails to account for the considerable lag between the rise of global trade and the collapse of European imperialism. Beliefs about the necessity of "trade with rule" doubtless varied across the population as well as over time, but the shift toward free trade in the mid-nineteenth century does not seem to have disinclined the British populace to put away the "big stick" for quite some time. We credit instead

    England's shrinking relative capacity advantage as the rest of Europe had closed the franchise gap.

    Reluctant Regimes: Small Franchise, Multiple Checks

    If a large franchise relative to neighbors is empowering, and the absence of checks is enabling, regimes with a small relative franchise and institutional checks, we might expect, would be militarily timid. Machiavelli thought so. He urged his native Florence to adopt the vigorous model of Republican Rome rather than the more narrowly franchised and sedate Sparta or Venice, both of which were popular models of republican government in his day, because he felt their military passivity would be no match for the looming threat of French and Spanish invasion.24

    Though Sparta managed to defeat Athens with the help of its subjugated allies, and

    though Venice defended itself successfully for centuries with mercenary armies, these regimes were known as status quo rather than expansionary powers.

    Early modern Poland affords perhaps the most extreme example in historical

    memory of the pacifying effects of institutional checks. Poland from the four teenth to the eighteenth centuries, like Venice, was an aristocratic republic in which

    approximately 5 to 10 percent of the adult male population shared political voice. The Polish commonwealth dated to the early middle ages when mounted warriors succeeded in pushing against the Turks to the south; Swedes, Lithuanians, and Rus sians to the north; and horseback tribes from the east; and in 1374, the warrior

    nobility institutionalized their diffuse decision-making system by stating explicitly that their king might levy taxes only with the permission of the Sejm, the legisla tive body of nobles. In this respect Poland resembled feudal realms elsewhere in

    Europe. But Poland resisted the trend that took hold in the rest of Europe in the sixteenth century toward absolutism in which a powerful monarchy had the author

    ity to raise increasingly larger armies for war. Although Poland's location against the eastern steppes and north of the Ottoman empire gave the cavalry-based war

    fare of the feudal nobility a certain enduring logic, the growth of centralized mon archies with large, infantry-based armies on its western and northern borders

    caught Poland short. The liberum veto, in which each Sejm legislator possessed a veto on all legislation, deprived successive Polish kings of the resources with which to raise a large peasant army. In the fifty-five diets from 1652 to 1764, forty-eight were "exploded" by the negative vote of a single deputy. This aristocratic

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 19

    obstructionism repulsed all efforts at institutional reform, including Rousseau's famous recommendation to increase the franchise, abolish the liberum veto, and

    strengthen the executive (Rousseau 1772). The result is that the Prussians and Russians helped themselves to large portions of poorly defended Polish territory in 1772, 1793, and 1795 (Lukowski 1999; Fedorowicz 1982).

    A Highly Checked Inclusive Regime: The United States

    Washington's Farewell Address is the classic statement of early American com

    mitment to "freedom from entangling alliances," but nowhere did this version of isolationism rule out expansion where convenient (Gilbert 1961, 1977, 15, 133). Hamilton, who is known to have put at least the finishing touches on Washington's most famous speech, had Palmerstonian views on trade but also thought it impru dent to challenge British maritime dominance. American expansion instead went

    westward, where the political and military ineffectiveness of Native American resis tance made for more tempting targets (Graebner 1985, 189). Jefferson's Louisiana Purchase in 1803 was a bargain at $15 million, even if Napoleon pocketed the prof its from land "he had no right to sell under the terms by which he had acquired it

    gratis from Spain" (Herold 1963, 311). The popular concept of Manifest Destiny, the Mexican War of 1846, the Spanish American War of 1898, and the Open Door

    Policy in China in 1900 were all of a piece in the very general sense that the Ameri can sphere of influence continued to expand, undeterred by American rhetorical

    commitment to?some would say a self-righteous moral obsession with?the right of democratic self determination.25

    Colonial assemblies began with a wide array of franchise arrangements (Wood ward 1902, 188-96), but as Alexander Keyssar (2000) has documented, American

    expansions in the franchise often accompanied military exploits. Most colonies had

    expanded their franchises during the Revolution, and Franklin was of the view that American soldiers had fought harder than British soldiers on account of being citizens with a political voice. At the Constitutional Convention Franklin opposed

    Morris's proposal for a national property qualification on suffrage rights on precisely these grounds: that the nation needed the full support of the populace (Keyssar 2000, 15). Hamilton was one of the few to favor giving slaves their freedom in exchange for joining the Continental Army, and gave an instrumentalist reason in addition to the moral one for doing this: "It should be considered that if we do not make use of them in this way, the enemy probably will..." (as quoted in Chernow 2004, 122). The South Carolina legislature rejected the slave enfranchisement plan, and the Northern states permitted the Southern states autonomy on the issue.

    Although states retained control over the franchise under the Articles of Confed eration, the Constitution gave the right to vote in national elections to those who could vote for the "most numerous Branch of the state legislature" (Keyssar 2000, 21). In Rakove's (1996, 214) view, "The need to mobilize an entire people for the

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 20 Journal of Conflict Resolution

    struggle encouraged new constituencies and segments of the population to gain a

    political voice, whether by correcting existing inequities in apportionment or by inspiring calls for an enlarged suffrage." Virginia abandoned its property require

    ment during the War of 1812 (Pole 1966, 18); in other states the franchise was broadened over the next few decades in the context of partisan competition for additional votes. Blacks finally won the right to vote during the Civil War, in

    part, no doubt, because their military contribution was valued, and the Nineteenth

    Amendment gave women the right to vote following World War I. There seems to be an inclination to draw in more voters to help with military conflict, but we also note that the expansion of the suffrage does not seem to bring in its wake a less

    aggressive foreign policy. The elimination of racial barriers to suffrage in the Civil

    Rights era took place in a cold war environment just as the United States's contain ment policy led to military intervention in South East Asia, though, to be sure, there was also a decline in voter turnout in this period.

    At least in some instances, expansion of territory or political influence has been

    stymied by American institutional fragmentation (Howell and Pevehouse 2007). Depending on the distribution of costs and benefits associated with the use of

    power, heterogeneous interests might be inclined to exercise veto points built into American political institutions; and even if their use is not often observed to block

    military aggression, their influence is likely to be incorporated in anticipation of their use. One of the few episodes in which the veto was not incorporated in advance was in 1846 when the Senate explicitly opposed military expansion in the Pacific Northwest favored by President Polk and the House of Representatives. The Senate opted for a compromise with Britain establishing the border between

    Oregon and Canada at 490 rather than to give military muscle to the "540 40' or

    Fight" pledge on which Polk had campaigned for the presidency (Nevins 1929; Merk 1967; Nelson 1981). That the very same Senate had agreed to fight Mexico for Texas and California suggests that the objection was to a bruising war with rela

    tively "republican" England rather than to war per se. The perceived probability of

    winning (or more generally, the expected utility of aggression) is a constraint that

    applies to all states' calculations about whether to fight, and the mobilizing capa city of another democracy is a special consideration in that context. Low-cost wars

    are not unpopular with either democratic masses or autocratic subjects.

    5. Medieval and Renaissance Italian

    City-states: A Statistical Test

    Italian city-states provide an intriguing setting in which to weigh Kant's insights about republican caution against Machiavelli's belief in democracies' fighting capacity. Italian city-states, on account of their rich variation in institutional forms

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 21

    Table 1 Political Variables and the Territorial Size of Italian City-States

    Independent Variable Estimation Results

    16,811.7 Constant (535.4542)

    0.000***

    -1,595.993 Number of veto points (299.9473)

    0.000*** 851.0023

    Popular rule (648.0444) 1.31

    N 1,463 R2 .0191

    Note: Cells contain parameter estimates, standard errors, and/?-values.

    Statistically significant at a < .001.

    including their elements of direct democracy, provide a potentially fruitful testing ground for the effects of popular voice and elite checks.

    Our first challenge was to find a reasonable way to capture our analytical vari

    ables of interest, popular check for political inclusivity, and number of vetoes for

    republican checks. For political inclusivity, we coded as 1 each city-state for every year in which a broad segment of the populace, defined at about 30 percent of

    males, had a voice in government. We relied heavily on the judgment of historians for whether a city-state had a popular check in a given year and coded these into

    "Histograms of Popular Voice," as shown in Appendix A. For republican checks, we attempted to measure the number of elite institutional vetoes with a role in

    policy making, particularly on matters of national security. Italian city-states varied

    in the number of institutional checks, from 1 in the case of a podesta or prince, to 5 or more when numerous councils shared governance of the city. Appendix B pro vides our list of the number of vetoes for each city-state, with constant coding between the dates listed.

    Finally, we ran simple regressions to see the relationship between these insti

    tutional variables, on one hand, and territorial size, on the other (see Table 1). Appendix C describes how we measured territorial size. The regressions indicate that institutional vetoes had a negative effect on territorial size, as Kant would have expected. This effect is statistically significant, suggesting that it is unli

    kely to have been generated randomly. Popular rule produces weaker statistical results but leans toward the positive effect that Machiavelli argued. It may be difficult to be more precise than this, for the years of popular rule are few even as we have defined it, and no one would claim that a 30 percent male suffrage is

    genuinely popular.

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 22 Journal of Conflict Resolution

    Table 2 Fixed Effects Model with Decade Dummies and a One-Year Lag

    Independent Variable Estimation Results

    10,736.890 Constant (864.342)

    0.000** -981.138

    Lagged veto points (299.885) 0.001***

    3,262.117

    Lagged popular rule (686.181) 0.000**

    268.283 Decade (30.514)

    0.000** N 1,463

    R2 .032

    Note: Cells contain parameter estimates, standard errors, and/?-values. ** Statistically significant at a < .005.

    *** Statistically significant at a < .001.

    Effects of the political variables on the ability or intention to expand may not be immediate, so in additional regressions, we included a one-year and a five-year lag

    (see Tables 2 and 3). The lag period is arbitrary, of course, because we do not have a theory about how long mobilizational capacity or veto inhibition should take, but we note that popular rule retains a positive coefficient and the number of vetoes

    retains a negative one in both models.26

    While our statistical analysis of the Italian city-states omits some potentially important variables for which we do not have data, we have no reason to believe

    that either of our political variables would be correlated systematically with apoliti cal mobilizational capacity such as population or GDP, since there are multiple paths to economic growth, at least in the short run. To the extent that political openness correlates positively with investment and growth in the longer run, it is consistent with the idea that political inclusivity provides the conditions for more full mobilization of the population's resources.

    Although these regressions are suggestive of an effect of political variables on

    fighting capacity and inclination, we note that even the more inclusive city-states had relatively low levels of political mobilization by today's standards. Italian city states were, in comparison to modern nation-states, mostly oligarchies of urban

    merchants fighting over which guilds should share in city government. Genoa was one of the few city-states to reach a threshold of inclusivity that unleashed the mobilizational advantage that Machiavelli hoped in vain to achieve in Florence.

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 23

    Table 3 Fixed Effects Model with Decade Dummies and a Five-Year Lag

    Independent Variable Estimation Results

    11,352.620 Constant (2,021.527)

    0.000***

    -1,228.600

    Lagged veto points (699.683) 0.080*

    3,331.721

    Lagged popular rule (1,591.784) 0.037**

    242.234

    Decade (67.462) 0.000***

    N 296 R2 .042

    Note: Cells contain parameter estimates, standard errors, and/?-values. * Statistically significant at a< .1.

    ** Statistically significant at a < .05.

    *** Statistically significant at

    a

  • 24 Journal of Conflict Resolution

    sometimes have chosen to fight in moments of passion. Whether the problem with direct democracies is the ease with which a majority under simple majority rules can expect to distribute the costs and benefits of war in its favor, or the lack of incentives to invest in information, or the vulnerability of mass assemblies to group psychology, it seems obvious that there are ample opportunities to blunder. Early modern Poland, at the other extreme, found itself tied in institutional knots it could not cut even in the face of grave national threat. We have argued for the usefulness

    of assessing political institutions on two dimensions: democratic inclusivity and

    republican checks.

    Machiavelli's notion of democratic capacity implies a hierarchy of mobiliza tional efficiency among a range of political systems, depending on how effectively they mobilize the resources of a given population. We leave it to future work to

    explore these ideas more fully and only suggest here that one way to interpret the rise of absolutism in early modern Europe is that monarchies represented broader social coalitions?including cities and peasants?than either the landed or mer

    chant aristocracies that dominated much of medieval Europe over preceding centu

    ries. This made them more efficient than aristocratic cavalry forces such as the

    Poles or the mercenary armies of Italian city-states, but monarchies in turn were

    out-mobilized by political regimes that were more inclusive still, as in the cases we have examined here.

    We recognize, of course, that rulers will give up political control only as a last resort and that they will try first other ways to mobilize societal resources that involve fewer political concessions. Religion, though a double-edged sword for the

    temporal ruler, competes formidably with democracy for mobilizational power. The kingdoms of Aragon and Castile united in 1492 in religious opposition to the Moslems in southern Iberia, giving Spain a territorial and financial basis for the

    largest national army in Europe for a time. Religious fires moved state boundaries in the rest of Europe as well, giving the greatest advantage to governments that

    could harness religion to the cause of nationalism, as in Holland and England. Even in France, where religious attachments and national interest often clashed in

    the Thirty Years War, the devastation from the religious wars between the Hugue nots and Catholic League made the provincial estates more ready to cooperate with

    the monarchy. Our argument about democratic inclusivity is a ceteris paribus one,

    and large powers of every stripe are likely to be expansionist because they have found some way to mobilize the resources of their populations more effectively than their neighbors. Even if nondemocracies may spend disproportionately more on defense to make up for their economic and other weaknesses (Goldsmith 2003), in the longer run nondemocratic powers may be frail if they have mobilized their

    populations on promises they cannot deliver.27

    The democratic peace literature has shown that democratic republics are more

    likely than nondemocracies to win wars they initiate, and we agree that this prob ably reflects the selection effect of politically inclusive regimes choosing to wage

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 25

    wars they think they can win, as well as their mobilizational capacity. Selection and mobilization, moreover, are likely to be synergistic in more than one way. Inclusive regimes generate better information about the risks and benefits of war, even as public enthusiasm for winnable wars translates into wider mobilization. In checked regimes, dubious military gambles have a greater chance of being win nowed out. Taken together, this suggests that inclusive regimes with institutional checks are likely to be the most formidable military powers, all else equal. As we have explained, expansion rather than victory in war provides the better measure of these effects. From the few cases available from the historical record, it seems that direct democracies such as Athens might have become embroiled in more "unpro ductive" wars than representative systems, and these rash acts may have impeded rather than advanced their cause.

    We return, in closing, to our conclusion that even highly checked republics may be aggressive if the opportunity for gain is tempting enough and that we would do well not to excuse the aggressive behavior of democratic republics as necessary acts of preemptive self-defense. Kant and neo-Kantian liberals are no doubt right to stress supplemental incentives for peacefulness such as trade and multilateral treaties.

    Appendix A

    Histograms of Popular Checks

    Siena: Popular checks

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 26 Journal of Conflict Resolution

    Florence: Popular checks

    Pisa: Popular checks

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 27

    Genoa: Popular checks

    i??* :!?": ': "" M

    Milan: Popular checks

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 28 Journal of Conflict Resolution

    Appendix B Vetoes

    Venice: Number of Veto Players

    1032: 2 (Doge no longer allowed to nominate colleagues and successors) 1223: 3 (Doge further limited by a Signoria) 1229: 4 (Senate of sixty chosen by the Great Council) 1310: 2 (Council of Ten streamlined foreign policy decisions) 1582: 4 (Powers of the Ten were curtailed)

    Florence: Number of Veto Players

    1261-66: 1 (Guelf) 1266-68: 2 1268-79: 1 (Ghib) 1282: 5 (Guild government with signoria, two colleges, Gonfalonieri and good

    men) 1326-28: 1 (Charles of Calabria) 1328-41: 5 (Guild government) 1342-43: 1 (Walter of Brienne) 1433: 2 (Albizzi had Cosimo arrested and banished into exile for ten years) 1434: 1 (Signoria was drawn that brought Medici back) 1458: 1 (Signoria was drawn that was favorable to the Medici) 1452: 1 (drawing Signoria by lot was suspended until 1468) 1478: 1 (Pazzi rebellion, April Blood, Lorenzo nearly assassinated) 1494: 2 (Savonarola, Great Council) 1502: 2 (Soderini, the Ten) 1512: 1 (Medici restoration) 1527: 5 (republican restoration) 1530: 1 (Medici)

    Siena: Veto Players

    1238: 3 (podesta, The Twenty-Four, council of the bell) 1287: 3 (podesta, The Nine, captain of the people on behalf of the general council) 1310: 2 (Nove without podesta, plus council) 1355:2 1368: 2 1389: 1 (Siena gave itself over to Visconti) 1404: 2 (independence under rule of the landed elite, some guild input)

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 29

    Milan: Number of Vetoes

    1198: 3 (Credenza of St. Ambrose got one of three officials) 1257: 2 (expulsion of nobility) 1277: 2 (Visconti, nobility) 1312: 1 1446-47: 5 (Ambrosian Republic) 1447:1 1499: 1 (conquest by France)

    Genoa: Number of Vetoes

    1266:2(Guelfvs. Ghib) 1269: 1 (Ghib podesta) 1306: 2 (Guelf captains) 1311: 1 (Henry VIII) 1313-31: 2 (civil war between Guelf and Ghib) 1335: 2 (Ghib) 1339: 1 (Ghib pop?lo, Boccanegra doge) 1344: 1 (Ghib noble represented by doge) 1356: 1 1394: 2 (Ghib noble) 1400: 2 (captains) 1421: 1 (Milanese rule) 1435: 2 (republic of nobles) 1436: 1 (pop?lo revolt) 1436: 2 (noble revolt) 1458: 1 (Charles VIII of France) 1461:2 1464: 1 (Milan) 1478: 1 (pop?lo and doge) 1487: 1 (Milan) 1499: 1 (France acquires Milan, Genoa) 1506: 2 (pop?lo, doge, eight tribunes) 1507: 1 (France) 1528: 2 (aristocratic republic)

    Pisa: Number of Vetoes

    1254: 3 (popular revolt; potesta, consuls, 12 anziani) 1316: 1 (Donnovatica) 1365-1402: 2 (Signoria and anziani councils) 1399: 1 (Milan) 1406: 1 (Florence)

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 30 Journal of Conflict Resolution

    Appendix C

    Measuring Territorial Size

    Our research assistant, Bryan Gervais, used two computerized map programs, Centennia and ArcGIS (ArcMap) to measure the size of each city-state each year between a.d. 1250 and 1600. Centennia shows the map of Europe every year from a.D. 1000 to the present. To calculate the area each year, Bryan saved the years between 1250 and 1600 as a bitmap file. He then uploaded the bitmap files into ArcGIS, a measuring program. Each bitmap file then had to be rectified against an actual map of Europe to put the correct dimensions and coordinates in place. He did this (with help from the Yale Map Collections staff) by selecting various points on the bitmap file maps and lining them up with the same points on the accurate

    Europe map. Once rectified, he created layers over the maps to take measurements.

    He rectified the layers as well. The measurement process consisted of creating "features" on the layer; he created features created by selecting points around the area he wished to measure. The more complex the feature, the more points he

    selected. He used many points when creating each feature (at times upwards of one

    hundred) so that the feature was as close as possible to the area being measured.

    ArcMap provides an attribute table for every feature created, including the total

    shape area and shape length. He took the total area for each feature and divided it

    by 1,000/1,000 (1,000 squared) to get the area in km2. When changes occurred, the

    polygon features could be cut to get the new number. For example, if a piece of Milan was successfully invaded, a cut would be made on the invasion line and the invaded territory area could be subtracted. For expansion, an ArcGIS tool called

    autocomplete polygon could be used to outline where the city-state expanded instead of creating a whole new feature and having lines overlap.

    The data include the acquisition and loss of Genoese and Venetian colonies in the Mediterranean.

    Appendix D

    Regressions of Italian City-State

    1. Siena

    Regression with robust standard errors

    Number of observations = 351

    F(2, 348) = 171.12 Prob >F =.0000

    /^-squared = .3640

    Root MSE = 5,118

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 31

    Robust 95 Percent Area Coefficient Standard Error t P>\t\ Confidence Interval

    No. of veto players -369.9494 639.8774 -0.58 .564 -1,628.463 to 888.5641

    Popular checks -7,843.5 500.6189 -15.67 .000 -8,828.12 to -6,858.881 _cons 10,525.39 1,359.986 7.74 .000 7,850.568 to 13,200.22

    2. Florence

    Regression with robust standard errors

    Number of observations = 320

    F(2, 317) = 414.19 Prob > F =.0000 R- squared

    = .3181

    Root MSE = 4,798.7

    Robust 95 Percent Area Coefficient Standard Error t P>\t\ Confidence Interval

    No. of veto players -2,175.347 102.9822 -21.12 .000 -2,377.962 to -1,972.732

    Popular checks 3,183.477 564.8519 5.64 .000 2,072.145 to 4,294.809 _cons 14,401.28 426.8918 33.74 .000 13,561.38 to 15,241.18

    3. Pisa

    Regression with robust standard errors Number of observations =157

    F(2, 154) = 204.14 Prob > F =.0000

    /^-squared = .6263

    Root MSE = 8,922.5

    Robust 95 Percent Area Coefficient Standard Error t P>\t\ Confidence Interval

    No. of veto players 1,204.065 2,415.812 0.50 .619 -3,568.343 to 5,976.473

    Popular checks 22,296.15 3,759.151 5.93 .000 14,896.99 to 29,722.3 _cons 5,626.751 3,819.733 1.47 .000 -1,919.985 to 13,172.59

    4. Venice

    Regression with robust standard errors Number of observations = 336

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 32 Journal of Conflict Resolution

    F(l, 334) = 34.89 Prob > F =.0000

    ^-squared = .0827

    Root MSE = 21,397

    Robust 95 Percent Area Coefficient Standard Error t P > \t\ Confidence Interval

    No. of veto players -16,313.55 2,761.799 -5.91 .000 -21,746.26 to -10,880.84 _cons 93,045.74 8,995.892 10.34 .000 75,350.01 to 110,741.5

    5. Milan

    Regression with robust standard errors

    Number of observations = 324

    F(2, 321) =13.14 Prob > F =.0000

    /^-squared = .2670

    Root MSE= 12,859

    Robust 95 Percent Area Coefficient Standard Error t P>\t\ Confidence Interval

    No. of veto players -18,478.27 5,455.155 -3.39 .001 -29,210.65 to -7,745.902

    Popular checks 1,744.163 2,542.178 0.69 .493 -3,257.271 to 6,745.596 _cons 48,719.76 6,186.05 7.88 .000 36,549.44 to 60,890.08

    6. Genoa

    Regression with robust standard errors

    Number of observations = 311

    F(2, 308) = 34.00 Prob > F =.0000 R- squared

    = .1133

    Root MSE = 2,318.8

    Robust 95 Percent Area Coefficient Standard Error t P>\t\ Confidence Interval

    No. of veto players -893.7379 325.5515 -2.75 .006 -1,534.324 to -253.1515

    Popular checks 1,551.493 196.0621 7.91 .000 1,165.702 to 1,937.283 _cons 13,454.1 433.9163 31.01 .000 12,600.29 to 14,307.92

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 33

    Notes

    1. It might be more accurate to use the term "republic" in place of "democracy" as most of the

    regimes to which our theories apply are not democracies in any strict (Athenian) sense of the term since the people have no direct part in government. But "democracy" is so entrenched in popular usage and

    "republic" as a term is so freighted with many other meanings that it seems best to stick with the

    accepted nomenclature rather than substituting one contested and complex usage for another. As we discuss below in more detail, we take seriously the differences between ancient, direct democracies and

    modern, representative ones. Another difference besides representation is that ancient democracies had small citizenries compared to residents on account of excluding women and slaves. But for our purposes here, the smaller citizenry does not make ancient democracies incomparable to modern ones since the men who voted were also those who bore arms.

    2. We use the term "republican" here in the way James Madison and Immanuel Kant did in the

    eighteenth century to mean a regime whose rulers are chosen by elections and in which the legislative and executive powers are separated in some way. Some prefer to call such regimes by other names: mixed governments, perhaps.

    3. Hansen (1991) demonstrates that the number of institutional checks increased at the end of the fifth century b.c., but it is doubtful that these restraints limited decisions to undertake military adven tures. In any case, the new restraints were democratic and differed mostly in the institutional way they could be expressed.

    4. The worry here is that even with panel data, most of the variation in the data might be cross sec tional rather than intertemporal. To some extent, fixed effects specifications can correct for some of this as long as the interaction effects are not pervasive. Russett (1995) has also urged the studying of conflict over the lifetime of each dyad. See also Diehl and Goertz (2000, 108). To meet the possible objection that ancient direct democracies and modern representative democracies are incomparable entities that do not belong in the same time series, we unpack regimes into (Kantian and Machiavellian) components that can be analyzed separately.

    5. Russett and Antholis (1992), Russett (n.d.), and Sobek (2003) are exceptions. To meet the possi ble objection that ancient direct democracies and modern representative democracies are incomparable entities that do not belong in the same time series, we unpack these regimes into (Kantian and Machia

    vellian) components that can be analyzed separately. 6. Bueno de Mesquita et al. (2003, 247) also make this point, but we think it is a fundamental meth

    odological point that deserves more attention. 7. This is a standard game-theoretic point about conflict in complete information settings. James

    Fearon (1995) summarizes this nicely and explains why, nonetheless, rational actors might choose to

    fight. States may defend themselves if attacked (Rousseau et al. 1996), to be sure, but they may also make concessions rather than to face the onslaught in the first place.

    8. In all these states, the popular influence was more or direct as in the classical cities. The grand councils of these cities were made up of the enfranchised people and might be wider or narrow cross

    sectionally or over time. For example, Venice's council shrunk from the late thirteenth century and was

    always relatively small. 9. In Kant's own words, from his 1786 essay on the "Perpetual Peace," "The reason is this: if the

    consent of the citizens is required in order to decide that war should be declared (and in this constitution it cannot but be the case), nothing is more natural than that they would be very cautious in commencing such a poor game, decreeing for themselves all the calamities of war. Among the latter includes: having to fight, having to pay the costs of war from their own resources, having painfully to repair the devasta tion war leaves behind, and, to fill up the measure of evils, load themselves with a heavy national debt that would embitter peace itself and that can never be liquidated on account of constant wars in the future" (as quoted in Reiss 1991).

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • 34 Journal of Conflict Resolution

    10. "Every form of government that is not representative is properly speaking without form, because one and the same person cannot be at the same time the legislator and executive of his will" (Kant, "Per

    petual Peace," as quoted in Reiss 1991, 114). Furthermore, "If the form of government is to cohere with the concept of right, it must include the representative system, which is possible only in a republican form of government and without which... government is despotic? None of the ancient so called republics were aware of this, and consequently they inevitably degenerated into despotism..." (ibid., 115).

    11. The democratic peace theory literature is vast, though it divides roughly into approaches that link

    peacefulness of democracies to electoral accountability mechanisms that internalize the costs of war

    (Bueno de Mesquita et al. 1999, 2003, 2004) and those that stress the importance of shared values and interests borne of democratic culture, trade, and diplomacy (Doyle 1983, 1986a, 1986b, 1997; Gartzke

    1998; Oneal and Russett 1999; Russett 1993, 1995, n.d.; Weart 1998). Both strands, however, draw their

    primary empirical support from the dyadic and war-based finding that democracies are less likely to fight each other than are democracies to fight nondemocracies or nondemocracies are to fight each other. Rousseau et al. (1996), Chan (1997), Gelpi and Griesdorf (2001), and Reiter and Stam (2002) provide insightful reviews of the democratic peace literature and its findings.

    12. It is probably wrong to think of Aristotle's mixed constitution as a system of checks; rather, he wanted to assure that the different parts of the city?the rich and poor?both had a role in government. The mixed constitution is probably better seen as a form of compromise or a balance between the parts of the city.

    13. Counting veto players affords one possibility, though not all checks have an equivalent effect on

    policy. 14. Machiavelli played the major role in creating a noncitizen militia in Florence in the context of

    the war against Pisa early in the sixteenth century. His reasons for this are well known: he opposed Flor ence's use of mercenary troops and thought that the best alternative was to rely on a conscripted militia. But we doubt that his leadership in this cause should be seen as an endorsement of the use of subjects rather than citizens in defense of the city. It seems much more likely that he thought that he could not

    hope to get Florentines to agree to citizen conscription and settled for a compromise. 15. Machiavelli states his view of the military power of political inclusiveness in Discourses, "All

    things considered, therefore, it is clear that it was necessary for Rome's legislators to do one of two

    things if Rome was to remain tranquil like [Venice and Sparta]: either to emulate the Venetians and not

    employ its plebs in wars, or like the Spartans, not to admit foreigners. Rome did both these things, and,

    by doing so, gave to its plebs alike strength, increase, and endless opportunities for commotion. On the other hand, had the government of Rome been such as to bring greater tranquility, there would have ensued this inconvenience that it would have been weaker, owing to its having cut off the source of

    supply which enabled it to acquire the greatness at which it arrived, so that, in seeking to remove the causes of tumults, Rome would have removed also the causes of expansion. So in all human affairs one

    notices, if one examines them closely, that it is impossible to remove one inconvenience without another

    emerging. If, then, you want to have a large population and to provide it with arms so as to establish a

    great empire, you will have made your population such that you cannot now handle it as you please. While, if you keep it either small or unarmed so as to be able to manage it, and then acquire dominions, either you will lose your hold on it or it will become so debased that you will be at the mercy of anyone who attacks you" (Machiavelli 1674, 1,6). See also Doyle (1997) and McCormick (1993).

    16. Machiavelli does not take sides, so to speak, in the current debate among democratic peace theor ists (Bueno de Mesquita et al. 2004; Reiter and Stam 2007) as to whether democratic mobilization works

    through selection effects (democracies choose wars they can win) or through more thorough marshalling of materials and effort. The current debate draws an unnecessarily sharp line between two complemen tary mechanisms if inclusive regimes tend to generate better information about which wars are worth

    fighting. In any case, Machiavelli's point is general enough to encompass both: "One finds that cities in which the populace is the prince, in a very short time extend vastly their dominions much more than do those which have always been under a prince; as Rome did after the expulsion of the kings, and Athens

    This content downloaded from 192.197.128.19 on Fri, 27 Sep 2013 11:18:38 AMAll use subject to JSTOR Terms and Conditions

  • Ferejohn, Rosenbluth / Warlike Democracies 35

    after it was free of Pisistratus. This can only be due to one thing: governm


Recommended